Short Term Load Forecasting for Uttarakhand using neural network and time series models

Accurate Load Forecasting is crucial in Power System operation and planning both for Regulated and De-regulated electricity markets. Short Term Load Fore-casting (STLF) is a tough job to handle because of the nonlinear behaviors of the electrical loads, climatic conditions. There is always a need for the sophisticated methods for STLF due to its complexity. This paper presented three models with neural network for both hour basis and day ahead prediction. The past load data as well as the weather data were collected and clustering technique is used for cleansing the data. The whole purified data is divided into three modules: Training data, Testing data and Validation data. The three NN models proposed were trained and tested by using the corresponding datasets. The presented models predicted the electric load, both daily and hourly, with high accuracy. The models proposed in this paper have been simulated using data obtained from State Load Dispatch Centre, Dehradun for the duration August 2011 to March 2013. The models are tested, results are compared with the actual load and Mean Absolute Percentage Error (MAPE) will be calculated.